Advances in Management & Applied Economics, vol. 5, no.5, 2015, 55-70
ISSN: 1792-7544 (print version), 1792-7552(online)
Scienpress Ltd, 2015
The Exchange Market Contagion in an Asymmetric
Framework before and after Abenomics
Chien-Chung Nieh1 and Hsun-Fang Cho2
Abstract
This study analyzes the contagion caused by the currency depreciation war in a multivariate
time varying asymmetric framework, focusing on countries competing with Japan for trade
during the Abenomic period. We employed not only the linear models of Engle and Granger
(1987), and the Johansen (1988) co-integration models but also the Enders and Siklos (2001)
asymmetric threshold co-integration model to investigate whether the contagion effects
between Japan’s exchange rate market and the exchange rate markets of major countries in
completion with Japan for export markets before and after Abenomics for the period 20112014 existed. The empirical evidence confirms a contagion effect particularly in Asian
countries where there is export competition with Japan with the exception of South Korea
during Abenomics. The contagion of the Japanese yen depreciation is not transmitted to
Australia, the Euro zone (France, Germany, Italy, and Netherlands), Qatar, Saudi Arabia,
and USA in competitive trade with Japan. We can apparently find the effect of yen
devaluation only occurred in the region of Asia close to Japan and does not spread Europe
and America. In general, our results support the contagion phenomenon for Abenomics.
Nevertheless, the effect of the contagion is regional not global.
JEL classification numbers: C32, F31, G01, G15
Keywords: Abenomics, Contagion, Currency depreciation war, Momentum threshold
autoregressive (M-TAR)
1
Introduction
Globalization and deregulation of financial markets promotes prosperity in the global
economy. The trend of trade liberalization has also led to more frequent financial dealings.
However, the impact of a shock to the finance, economy, and the politics in one country
1
2
Department of Banking and Finance of Tamkang University, Taiwan.
Department of Banking and Finance of Tamkang University, Taiwan.
Article Info: Received : July 2, 2014. Revised : July 31, 2015.
Published online : October 15, 2015
56
Chien-Chung Nieh and Hsun-Fang Cho
often crosses to countries in the same region or countries in other regions, which result in a
regional or a global economic crisis, such as the Mexico peso crisis in 1994, the Asian
financial crisis in 1997, the Russian financial crisis in 1998, the U.S. Dot-com bubble in
2000-2002, and the global financial crisis in 2007-2008. Currency crises have occurred
repeatedly over the past twenty years. Researcher has investigated how attacks of
speculative behavior create a currency crisis in one country; the market volatility tends to
spread to other countries in the specific region and elsewhere. Researchers and academic
institutes have proposed that exogenous events, and shock transmission explains this
phenomenon, generally referred to as contagion. The recent crisis-contagion theory is
ardently discussed by the academic authorities and policy makers, especially in light of the
frequent economic and financial crises all over the world.
As we all know, the Japanese economy has been depressed for 26 years since the asset
bubble burst in 1989 (The Nikkei 225 index reached its highest point at 38,913). The
Japanese economy required a stimulus to escape from this pattern of long-term sluggish
growth. The Liberal Democratic Party (LDP) overwhelmingly won a general election
which took place in Japan on December 16th, 2012. Abe Shinzo regained the power to
govern as Prime Minister on December 26th, 2012. His prescription for economic
reactivation was referred to as Abenomics which is a new economics policy regime and a
new term – Abenomics – that is used to refer to the three pillars or arrows for the Japanese
economy and economic policy. The first arrow is the unconventional monetary policy; the
second arrow is the expansionary fiscal policy and the third arrow is the economic growth
strategies. The Japanese government tried to revive its economy through implementing bold
economic policies that will pull its economy out of prolonged deflation. The Abenomics
policy led to a dramatic weakening of the Japanese yen. According to Figure 1, the yen
became about 25% lower against the U.S. dollar in the second quarter of 2013 compared to
the same period in 2012, with an extremely loose monetary policy being followed. The
Bank of Japan adopted a policy of quantitative easing aimed at creating a sharp depreciation
of the yen. This caused Japan’s trading competitors to become afraid that their exports are
uncompetitive. These countries joined together with a policy of competitive devaluation of
currencies in order to maintain export competitiveness. Japan triggered a currency war
causing contagion in competitive devaluation of currencies. Understanding this issue is one
of the purposes of this study by investigating the effect of the sharp depreciation of the yen
during the period of Abenomics. Therefore, the general discussion about contagion explores
how to define contagion t in the first and then considers how to measure contagion.
In terms of the definition, generally, the influential definitions of contagion refer to a
significant increase in cross-market linkages after a shock to one country or group of
countries and a cross-market correlation. Otherwise, where there is a continued market
correlation at high levels, this is considered to be “no contagion, only interdependence”
(Forbes and Rigobon, 2002, Bekaert et al., 2005). Commonly, contagion refers to the spread
of financial disturbances from one country to others. The literature on financial contagion
has literally exploded since the publication of the paper by Forbes and Rigobon (2002)
stimulated extensive debate and discussion. Much of the extant empirical literature on
contagious currency crises stresses the phenomenon of regional contagion. Lee and Kim
(1993), Forbes and Rigobon (2002), Dungey, Fry, González-Hermosillo, and Martin (2006),
Lucey and Voronkova (2008), and Arouria, Bellalahb, and Nguyenc (2009) about the
transmission effect and the contagion effect were based on the backgrounds of several crises
since the late 1990s, such as those in Mexico (1994), Thailand (1997), Russia (1998), and
Argentina (1999). Calvo and Reinhart (1996) reported correlation shifts during the Mexican
Exchange Market Contagion in Asymmetric Framework before and after Abenomics
57
Crisis, while Baig and Goldfajn (1999) supported the contagion phenomenon during the
East Asian Crisis.
In the light of the need to quantify contagion, during recent last years, scholars have been
using more advanced techniques to measure contagion. For example, Patton (2006)
pioneered the study of time-varying copulas for modelling asymmetric exchange rate
dependence. Bartram et al. (2007) estimated time-varying copula dependence models for
17 European stock market indices, following the methodology of Patton (2006). In the
recently developed area of regime-switching copulas, Rodriquez (2007) modeled
dependence with switching-parameter copulas to study financial contagion and provides
evidence of changing dependence and asymmetry during the Asian and the Mexican crises.
Okimoto (2008) estimated regime-switching copulas for the US–UK pair and found that
the bear regime was better described by an asymmetric copula with lower dependence. In
terms of other empirical literature, Caporale, Cipollini, and Spagnolo (2005) modeled the
conditional variance by the application of both heteroskedasticity and endogeneity biases
and invented a common shock to deal with the omitted variable problem. They found the
existence of contagion within the stock markets in Hong Kong, Japan, South Korea,
Singapore, Taiwan, and Malaysia during the 1997 Asian Financial Crisis. The findings were
consistent with the crisis-contingent theories of stock market linkages. Bekaert, Harvey,
and Ng (2005) also reported that co-integration relationships did exist among the Asian
stock markets during the period of the 1997 Asian Financial Crisis, which demonstrated a
contagion effect. Li and Lam (1995), Koutmos (1998), and Chiang (2001) pointed out that
co-integration between stock markets was asymmetric; Wang and Lin (2005), Shen, Chen,
and Chen (2007), and Chang (2008, 2010) further employed the asymmetric co-integration
test for their empirical studies. This paper presents a theory for analyzing whether intercountry trade can be responsible for the transmission of a currency crisis, which has
important implications for understanding the empirical phenomenon in general and
possibly also its regional dimensions.
To investigate how the asymmetric adjustment phenomenon influences the contagion effect
or the transmission effect, we apply the asymmetric threshold co-integration method to
compare the transmission effect or the contagion effect from the Japanese exchange rate
market to the exchange rate markets of import and export countries trading with Japan in
the pre-Abernomics and during the Abenomic period; therefore, asymmetric adjustments
could exist in an upward status (positive impact) or a downward status (negative impact).
How do the two phenomena influence the transmission effects or contagion effects of the
exchange rate markets? Do different correlations, co-movement, interdependence, or
contagion effects exist during currency depreciation? These issues were seldom discussed
in previous literature; therefore, we decided to explore these problems by the asymmetric
threshold co-integration model. What is the impact of the Japanese yen depreciation on the
countries which have a trade relationship with Japan during the period of Abenomics? Is
co-integration strengthened during the great depreciation? The issue of the contagion effect
in some countries in Asia, Europe and America, which we have selected for this paper, is
carefully examined.
The structure of the paper is organized as follows. Section 2 presents data description and
the econometric method. Section 3 shows empirical results. Finally, concluding remarks
are stated in Section 4.
58
2
Chien-Chung Nieh and Hsun-Fang Cho
Data Description and the Econometric Method
2.1 Data and Variables
The study aims to investigate the asymmetric contagion effect of the Japanese exchange
rate on the exchange rate of the major export and import countries trading with Japan3 such
as European countries including Germany, France, Netherlands, and Italy, Asian countries
consisting of China, South Korea, Thailand, Saudi Arabia, Taiwan, Malaysia, Hong Kong,
Indonesia, Singapore, United Arab Emirates, and Qatar, North America including the
United States, Canada and Australia. Because the four European countries have adopted the
euro (€) as their common currency and sole legal tender, the currencies of these countries
are identical utilizing the euro (€) in this study. The exchange rates are quoted in the paper
employs the price per unit of US dollar expressed in the currency of the target country. Here,
the US dollar is called the "Fixed currency", while other country’s currencies are referred
to as a "Variable currency". The exchange rate of the Unite States adopts the U.S. Dollar
Index4. All of the variables in this study are taken from logarithms.
In order to sufficiently investigate the long equilibrium relationship of variables, this
research utilizes sample data over four years. Therefore, daily data arranged in 5-day weeks
is selected for this dissertation from 1st January2011 to 31st December 2014 and downloaded
from the Datastream database for exchange rates of nineteen countries including sixteen
currencies. A total of 1,043 observations are obtained for each variable, which are utilized
to analyze the extent of co-movements and contagion effect of exchange rare depreciation
from Japan to other countries. To account for the impact of Abenomics on the exchange
rates of countries importing and exporting to Japan, The study conducts a process of
analysis observing all variables over the same period. Table 1 displays the descriptive
statistics showing the exchange rates of all the countries investigated.
The study wants to investigate the influence of the implementation before and after
Abenomics. The periods before and after Abenomics need to be defined for observing
whether there exists significant difference between two periods. According to Prime
Minister Yoshihiko Noda’s announcement, the House of Representatives was dissolved by
Prime Minister Yoshihiko Noda on November 16th, 2012 as so this marks the cut off point
for the pre-Abenomic period. After that, the Japanese exchange rate started to dramatically
depreciate as illustrated in Figure 1. The period before Abenomics is defined as being from
January 1st, 2011 to November 16th, 2012 to provide a total of 490 daily observations. The
period of Abenomics is defined as being from November 19th, 2012 to December 31st, 2014
with 553 days of daily data. We, therefore, compared the estimated results for the different
periods.
All variables are transformed into natural logarithms namely 𝐿𝐸𝑅𝑖,𝑡 = 𝑙𝑛𝐸𝑅𝑖,𝑡 . Where ER
is the abbreviation for exchange rate. t denotes the time period of observation. i represents
observed the countries. The equation can be expressed as:
3
For more details regarding the information, the reader is referred to the website of the observatory
of economics complexity: Source: https://atlas.media.mit.edu/en/profile/country/jpn/
4
The US Dollar Index (USDX, DXY) is an index (or measure) of the value of the United States dollar
relative to a basket of foreign currencies, often referred to as a basket of US trade partners' currencies.
Source: https://en.wikipedia.org/wiki/U.S._Dollar_Index
Exchange Market Contagion in Asymmetric Framework before and after Abenomics
59
∆𝐿𝐸𝑅𝑖,𝑡 = (𝑙𝑛𝐸𝑅𝑖,𝑡 − 𝑙𝑛𝐸𝑅𝑖,𝑡−1 ) × 100
where ∆𝐿𝐸𝑅𝑖,𝑡 , 𝑖 = 1,2 … 16 represents the variables in the first difference.
(1)
125
Yen / US
115
105
95
85
75
1/3/2011
1/3/2012
1/3/2013
1/3/2014
Figure 1: The graph of Japanese yen from 2011 to 2014
Table 1: Summary statistics for return on exchange rates
Max
Std.
Skewnes Kurtosi
Mean
Min.
J-B
.
Dev.
s
s
3.71 Japan
0.037
0.586
0.655
7.763 1,059***
0
2.327
4.10 Australia
0.021
0.666
0.283
6.454 532***
8
3.277
2.95 Canada
0.015
0.467
0.165
6.228 457***
5
1.962
0.49 China
-0.006
0.100
-0.101
7.947 1,064***
9
0.586
2.10 Euro
0.010
0.550
0.064
4.519 101***
4
2.307
0.14 Hong Kong
0.027
-0.301
13.547 4,845***
0.0002 9
0.222
3.13 Indonesia
0.031
0.370
0.457
11.509 3,180***
9
1.726
1.68 Malaysia
0.013
0.402
-0.187
6.293 477***
1
2.648
0.07 Qatar
0.000
0.010
-0.052
19.968 12,501***
8
0.072
0.07 447,455**
Saudi Arabia
0.0001
0.005
-0.136
104.519
3
0.076
*
2.14 Singapore
0.003
0.359
0.368
9.151 1,666***
0
2.290
2.79 South Korea
-0.002
0.472
0.606
6.092 479***
7
1.709
1.54 Taiwan
0.008
0.202
0.703
9.954 2,185***
0
0.851
0.008 1.47 0.306
-0.245
6.975 697***
Thailand
Obs.
1,04
3
1,04
3
1,04
3
1,04
3
1,04
3
1,04
3
1,04
3
1,04
3
1,04
3
1,04
3
1,04
3
1,04
3
1,04
3
1,04
60
Chien-Chung Nieh and Hsun-Fang Cho
United Arab
Emirates
US
2
0.01
0.000
1
1.86
0.013
6
2.267
0.002
0.011
0.431
1.861
0.343
0.182
3
1,04
10.946 2,762***
3
1,04
4.696 131***
3
Notes:
1. Std. Dev. denotes the standard error. Max. represents the maximum. Min. is the
minimum. Obs. is the observation
2. J-B denotes the Jarque–Bera normality test.
3. *** indicates significance at the 1%, respectively.
2.2 Econometric Method
The Engle and Granger (1987) threshold autoregressive (TAR) and momentum threshold
autoregressive (M-TAR) models for unit root testing to permit unit root tests within a
multivariate framework, and to allow for nonlinear adjustments were generalized by Enders
and Siklos (2001). They are better in picking up these asymmetries than the linear models
of Engle and Granger (1987) or the Johansen (1988) co-integration models. The
methodology involves enforcing two procedures where the first stage estimates a long-run
relationship based on:
𝑌𝑡 = 𝛼 + 𝛽𝑋𝑡 + 𝜇𝑡
∆𝜇𝑡 = 𝜌𝜇𝑡−1 + ∑𝑘𝑖=1 𝜙𝑖 ∆𝜇𝑡−1 + 𝜀𝑡
(2)
(3)
where 𝑌𝑡 and 𝑋𝑡 are two I(1) series, t denotes the time period, 𝜇𝑡 is a stationary random
variable that denotes deviation from the long-term equilibrium, assuming that a long-run
relationship exists. α , β , 𝜌 and 𝜙𝑖 are estimated parameters. 𝜀𝑡 is a white noise
disturbance term. 𝑘 is the number of lags.
In the first stage of estimating the long-term relationship between the variables, the
regression is normalized on the downstream variable 𝑌𝑡 . In the second stage, the residuals
𝜇𝑡 are utilized to enforce a cointegration test. The number of lags is chosen by the Akaike
information criterion (AIC), Schwarz Bayesian information criterion (SBC), or Ljung–Box
Q test, so that there is no serial correlation in the regression residuals. If the null hypothesis
of 𝐻0 : ρ = 0 is rejected, then the two variables are regarded as to be linearly co-integrated.
The above co-integration tests assume symmetric transmission. Enders and Siklos (2001)
declared a two-regime threshold co-integration method to permit asymmetric adjustments
in the co-integration test. To introduce asymmetry in the adjustment to the long-term
equilibrium, the adjustment process is dependent on a change in 𝜇𝑡 , (∆𝜇𝑡 ). The residuals
from Eq. (3) are utilized to evaluate a model of the form. The next stage requires estimation
of two parameters, 𝜌1 𝑎𝑛𝑑 𝜌2 in the following equations:
∆𝜇𝑡 = 𝐼𝑡 𝜌1 𝜇𝑡−1 + (1 − 𝐼𝑡 )𝜌2 𝜇𝑡−1 + ∑𝑙𝑖=1 𝛾𝑖 𝛥𝜇𝑡−𝑖 + 𝜀𝑡
where 𝐼𝑡 = [𝑇𝑡 , 𝑀𝑡 ], such that:
(4)
Exchange Market Contagion in Asymmetric Framework before and after Abenomics
1
𝑖𝑓 𝜇𝑡−1 ≥ 𝜏
𝑇𝑡 : TAR = {
0
1
𝑀𝑡 :M-TAR={
0
61
(5)
𝑖𝑓𝜇𝑡−1 < 𝜏
𝑖𝑓 ∆𝜇𝑡−1 ≥ 𝜏
(6)
𝑖𝑓∆𝜇𝑡−1 < 𝜏
where 𝜀𝑡 is a white-noise disturbance. 𝐼𝑡 is the Heaviside indicator function. 𝑙 is the lag
term which is again specified to interpret serial correlation in the residuals and it can be
chosen by the AIC, SBC, or Ljung–Box Q test. τ is the threshold value, which is priorly
unknown and has to be estimated. The threshold value is endogenously decided by using
the Chan (1993) grid search method to find out an estimate of the consistent threshold value.
The Heaviside indicator 𝐼𝑡 can be specified by two alternative definitions of the threshold
variable, either the lagged residual (𝜇𝑡−1 ) or the change of the lagged residual (∆𝜇𝑡−1 ).
Equations (4) and (5) denote the threshold autoregressive model (TAR). Equations (4) and
(6) represent the momentum threshold autoregressive model (M-TAR). Enders and Granger
(1998) found the M-TAR model was particularly valuable when adjustment was
asymmetric such that the series displayed more “momentum” in one direction than the other.
In the TAR model, the adjustment is modified by 𝜌1 𝜇𝑡−1 that 𝐼𝑡 = 𝑇𝑡 = 1 when the
residual according to equation (5) is above the threshold value 𝜏 and by the term 𝜌2 𝜇𝑡−1
that 𝑇𝑡 = 𝐼𝑡 = 0 when the residual is below the threshold value. In the M-TAR model, the
adjustment is modeled by 𝜌1 𝜇𝑡−1 that 𝐼𝑡 = 𝑀𝑡 = 1 when the residual according to
equation (6) is above the threshold value 𝜏 and by the term 𝜌2 𝜇𝑡−1 that 𝑀𝑡 = 𝐼𝑡 = 0
when the residual is below the threshold value.
Since there is generally no consensus about which specification should be used, it is
suggested that the proper adjustment mechanism is chosen through the model selection
criteria of AIC and SBC (Enders and Siklos, 2001). The model with the lowest AIC and
SBC will be utilized for further analyses. Insights into asymmetric adjustments in the
context of a long-term co-integration relationship can be acquired through two tests on the
coefficients estimated from the threshold co-integration equations. First, Enders and
Granger (1998) and Enders and Siklos (2001) concluded that in either case, under the null
hypothesis of no convergence, the F-statistic for the null hypothesis of 𝐻0 : 𝜌1 = 𝜌2 = 0
had a nonstandard distribution. Enders and Granger (1998) also found that if the series is
stationary, the least square estimates of 𝜌1 and 𝜌2 have an asymptotic multivariate
normal distribution. We test the null hypothesis of 𝐻0 : 𝜌1 = 𝜌2 = 0 for a co-integration
relationship, and rejection implies the existence of a cointegration relationship between the
two variables. Second, if the null hypothesis of no co-integration is rejected, it would enable
us to advance to further test for symmetric adjustment of the null hypothesis, which is
𝐻0 : 𝜌1 = 𝜌2 . We proceed with the asymmetric threshold cointegration test and symmetric
adjustment test by using the standard F-statistic. Rejection of the null hypothesis indicates
the existence of an asymmetric adjustment process between the two variables.
62
3
Chien-Chung Nieh and Hsun-Fang Cho
Empirical Results
3.1 Advanced Non-linear ESTAR5 Unit Root Test
The conventional linear unit root tests6 might have a lower power when they are applied to
a finite sample. In this situation, the advanced non-linear ESTAR unit root test is found to
be of great help provided that they permit an increase in the power of the order of the
integration analysis by allowing for asymmetric adjustment to the equilibrium level. The
results of the nonlinear unit root test are provided in Table 2. As can readily be seen from
the table, the results of the test suggest that the null hypothesis of the unit root is rejected
in the circumstance of the first difference at the 1% significant level for most variables of
all exchange rate markets in this paper.
Table 2: Results of the non-linear unit root test
Level
First difference
0.762(0)
-32.69(0)***
Japan
0.149(0)
-4.656(3)***
Australia
-0.064(0)
-2.397(5)**
Canada
-2.338(0)**
-33.05(0)***
China
-1.525(4)**
-9.589(5)***
Euro
-1.354(4)
-16.46(3)***
France
-1.264(4)
-11.10(5)***
Germany
-2.672(1)**
-29.41(0)***
Hong Kong
0.437(5)
-12.55(4)***
Indonesia
-1.376(4)
-16.54(3)***
Italy
-0.327(4)
-17.66(3)***
Malaysia
-1.293(4)
-11.85(5)***
Netherlands
-6.817(4)***
-17.64(6)***
Qatar
1.232(5)
-19.85(4)***
Saudi Arabia
-2.560(0)**
-34.32(0)***
Singapore
-2.247(3)**
-17.48(2)***
South Korea
-1.061(2)
-19.47(1)***
Taiwan
-1.368(1)
-31.19(0)***
Thailand
-18.53(0)***
-33.55(1)***
United Arab Emirates
-0.332(1)
-33.99(0)***
US Index
Notes:
1. *** , **and * denote significance at 1%, 5% and 10% levels, respectively.
2. The numbers in the parentheses are the appropriate lag lengths selected by MAIC
(modified Akaike information criterion) suggested by Ng and Perron (2001).
3. The simulated critical values for different K were tabulated in Kapetanios et
al.(2003) (Table 1 as of p. 364).
5
We employ the exponential smooth transition autoregressive (hereafter, ESTAR) unit root tests
proposed by Kapetanios et al. (2003).
6
The conventional linear unit root tests include the Augmented Dickey-Fuller (Dickey and Fuller,
ADF, 1981), test, Phillips, the Perron (Phillips and Perron, PP, 1988) test, and the KwiatkowskiPhillips-Schmidt-Shin (Kwiatkowski et al., KPSS, 1992) test.
Exchange Market Contagion in Asymmetric Framework before and after Abenomics
63
3.2 Engle-Granger Co-integration Test
We can conclude that all variables are integrated of the same order after running unit root
tests. The next task is to check for co-integration so the co-integration test is utilized to
analyze the data. This approach is executed to investigate the long-run equilibrium
relationship between variables. The results of the Engle-Granger co-integration relationship
between Japan and the exchange rates of other countries over the entire period, the period
of the pre-Abenomics, and the period of during-Abenomics, and the null hypothesis of no
co-integration is shown in Table 3. The results of the Engle-Granger ADF statistics show
that there are co-integration relationships between Japan’s exchange rate and the exchange
rates of Canada, Hong Kong, Qatar, and United Arab Emirates over the entire period.
Furthermore, the results show that there are co-integration relationships between Japan and
Australia, Canada, Hong Kong, Qatar, Saudi Arabia, United Arab Emirates during the preAbenomic period. As shown in the table, the results show that there are co-integration
relationships between Japan and Hong Kong, Qatar, and United Arab Emirates in the
during-Abenomics period. It can be seen from Table 3 that there is a significant increase in
the co-integration relationships between Japan’s exchange rate and the exchange rates of
Hong Kong and the United Arab Emirates around the time that Abenomics were applied;
this result does not support the contagion theory by Dornbusch et al. (2000) and Forbes and
Rigobon (2001).
Table 3: Results of the Engle-Granger test for co-integration
Australia
Canada
China
Euro
Hong Kong
Indonesia
Malaysia
Qatar
Saudi Arabia
Singapore
South Korea
Taiwan
Thailand
United Arab
Emirates
US Index
Entire period
Engle-Granger
ADF Statistic
-2.945(0)
-3.139(0)
-0.947(0)
-1.819 (0)
-3.207 (1)
-1.725(0)
-2.595(0)
-7.263(3)
2.593(10)
-2.818(0)
-2.279 (0)
-2.160(2)
-2.000(0)
-13.75(1)
Pre-Abenomics
Engle-Granger
ADF Statistic
-3.131(0)
-3.276(0)
-2.094(1)
-2.845(0)
-3.835(0)
-2.465(2)
-2.721(0)
-9.582(1)
-4.647(4)
-2.344 (0)
-2.241 (0)
-2.623(2)
-2.370(0)
-10.93(1)
During Abenomics
Engle-Granger
ADF Statistic
-2.362 (0)
-3.009 (0)
-1.022(0)
-1.055(0)
-4.231(2)
-1.457(0)
-2.502(0)
-5.973(1)
-0.302(12)
-2.757(0)
-1.245(0)
-2.589(0)
-1.968(0)
-12.35(0)
-1.752(1)
-2.822(0)
-0.781(1)
Notes:
1. The numbers in the parentheses are the appropriate lag lengths selected by
minimizing AIC.
2. The critical values of the Engle-Granger ADF Statistics are taken from Engle and
Yoo (1987).
3. *, ** and *** denote significance at the 10%, 5% and 1% significance levels,
respectively.
64
Chien-Chung Nieh and Hsun-Fang Cho
It is clear from Table 4 that the results of the Johansen maximum eigenvalue co-integration
test for the entire period are that there are co-integration relationships of one co-integrating
rank between Japan’s exchange rate and the exchange rates of Hong Kong, Qatar, and
United Arab Emirates. As can be seen from the table, there are co-integration relationships
of two co-integrating ranks between Japan and Hong Kong, Saudi Arabia, and United Arab
Emirates in the pre-Abenomics period. The evidence for the during-Abenomics is shown
in Table 4; the empirical evidence shows that Japan and Hong Kong, Qatar, and the United
Arab Emirates have cointegration relationships of one rank. According to the results, it does
not support the contagion theory by Dornbusch et al. (2000) and Forbes and Rigobon (2001).
Table 4: Results of the Johansen maximum eigenvalue co-integration test
Rank
Australia
Canada
China
Euro
Hong Kong
Indonesia
Malaysia
Qatar
Saudi Arabia
Singapore
South Korea
Taiwan
Thailand
United Arab
Emirates
U.S.
r=0
r≦1
r=0
r≦1
r=0
r≦1
r=0
r≦1
r=0
r≦1
r=0
r≦1
r=0
r≦1
r=0
r≦1
r=0
r≦1
r=0
r≦1
r=0
r≦1
r=0
r≦1
r=0
r≦1
r=0
r≦1
r=0
r≦1
Entire
period
Max-Eigen
statistic
13.76
4.057
9.874
3.110
3.532
1.881
5.346
3.364
26.77
3.162
6.431
2.941
9.461
4.563
68.63
3.212
9.388
0.367
7.980
3.258
5.771
3.428
5.495
3.793
6.400
4.785
149.0
3.192
5.823
3.592
PreAbenomics
Max-Eigen
statistic
11.78
4.747
11.71
4.930
5.402
4.014
10.62
4.685
19.31
5.751
7.468
6.680
9.693
3.391
62.41
4.881
41.25
5.038
8.443
4.337
7.155
3.597
10.32
2.546
6.390
3.851
81.88
4.892
11.39
4.721
During
Abenomics
Max-Eigen
statistic
14.91
4.235
11.93
3.514
4.421
1.056
11.24
0.0354
27.444
3.486
4.398
2.476
11.00
3.408
30.82
3.476
13.28
0.3259
12.59
2.876
11.82
1.728
12.53
2.340
6.593
5.143
74.33
3.408
11.96
0.024
3.3 Results of the Threshold Co-integration Test
Enders and Granger (1998) and Enders and Siklos (2001) proposed two models for the
Exchange Market Contagion in Asymmetric Framework before and after Abenomics
65
threshold co-integration test, namely, the TAR model and the M-TAR model. This study
adopts the M-TAR model. Enders and Granger (1998) suggested that when asymmetrical
adjustments occurred in the data series, the determination of the Heaviside indicator
function might also be decided by the first difference value of error correction term on the
period t − 1 (∆𝜀𝑡−1 ) . Boucher (2007) pointed out that the speed of convergence of
parameter estimation by using the M-TAR model would be faster than that of the TAR
model. Table 5 presents the results for the threshold co-integration relationship between the
exchange rate market of Japan and the exchange rate markets of observed countries in this
study. The null hypothesis of no co-integration (𝐹𝑐 ) and symmetric adjustment (𝐹𝐴 ) are
also shown in Table 5. The empirical evidence shows that the null hypothesis of no cointegration (𝐹𝐶 ) is rejected at the 10% significant level for the entire period, which
indicates the existence of long-run equilibrium relationships between Japan’s exchange rate
and exchange rates of other countries. What is more, the null hypothesis of symmetric
adjustment (𝐹𝐴 ) is rejected at the 10% significant level in the entire period except for
China, the Euro, and Malaysia, suggesting that there is a significant threshold of cointegration and asymmetric adjustment between the two variables considered. The null
hypothesis of no co-integration (𝐹𝐶 ) is rejected in the pre-Abenomics of Table 5 with the
exception of China, Indonesia. Furthermore, the null hypothesis of symmetric adjustment
(𝐹𝐴 ) is rejected, suggesting that there exists asymmetric adjustments between Japan and
Euro, Hong Kong, Qatar, Saudi Arabia, Singapore, South Korea, Taiwan, and the U.S..
Moreover, Japan has co-integration relationships (𝐹𝐶 ) with all observed countries during
the Abenomics period except for the Euro. In terms of asymmetric adjustment (𝐹𝐴 ), there
is an asymmetric relationship between Japan and other countries during the Abenomics
except Euro, Malaysia, Saudi Arabia, Taiwan. By further comparisons of the 𝐹𝐶 statistics
in the pre-Abenomics period and the during-Abenomics period shown in the (10) column
of Table 5, we have found that the co-integration relationships significantly increased
between Japan exchange rates and the exchange rates of observed countries in the study
except Australia, Euro, Qatar, Saudi Arabia, and South Korea. The result shows that there
is a “contagion effect” or “transmission effect” between the Japanese exchange rate and the
exchange rates of observed countries in the study, but there is only an “interdependence
effect” between Japan and Australia, Euro, Qatar, Saudi Arabia, and South Korea. Forbes
and Rigobon (2001) defined the contagion of the international financial markets as a
significant increase in cross market linkages or co-movement between one market and
others after a shock or during a crisis, and our results supported the “contagion effect”
between Japan exchange rate market and parts of exchange rate markets in the surveyed
countries in our study. In addition, by further comparisons of the 𝐹𝐴 statistics in preAbenomics and during-Abenomics in the (11) column of Table 5, we have found that the
asymmetry in the co-integration relationships has also significantly increased after the end
of Abenomics between Japan and most of the observed countries in the paper. The result
shows that a result of Abenomics was the quick transmission of massive amounts of
negative information between many exchange rate markets. This led to higher risk aversion
among international investors.
According to the empirical results, we find out that there are quite large differences between
utilizing the Enders-Siklos’ asymmetric threshold test (M-TAR), and the Engel-Grange and
Johansen co-integration tests with the Enders-Siklos test producing better results. It is
apparent from Table 3 and Table 4 that the co-integration relationships between the
Japanese exchange rates and exchange rates of surveyed countries during the Abenomics
66
Chien-Chung Nieh and Hsun-Fang Cho
period do not significantly increase. Some researched results demonstrate that the
relationships between international exchange markets should have a closer co-integration
correlation when the emergence of severe risk impacts the global economy. We are unable
to obtain results showing the co-integration degree increases when we apply the EngelGrange and Johansen co-integration tests because the hypotheses of the two co-integration
tests rely on a symmetric adjustment process. However, the result of the Enders-Siklos
asymmetric threshold co-integration shows that there is a significant increase in the cointegration correlation during the Abenomics period between the Japanese exchange rate
and the exchange rate of the surveyed countries in the study. The co-integration
relationships are regarded as a trend of mutual movement. After Abenomics, the cointegration relationships increased between Japanese exchange rates and the exchange rates
of Canada, China, Hong Kong, Indonesia, Malaysia, Singapore, Taiwan, Thailand, United
Arab Emirates, and the U.S.. The results also illustrate that exchange rate depreciation in
Japan caused contagion effect. That can also be called the phenomena of competitive
depreciation of exchange rates. However, there are only interdependent effects between
Japan and Australia, Euro, Qatar, Saudi Arabia, South Korea. During Abenomics, Asian
countries in competition with Japan to export adapted their currency devaluation to promote
competitive advantage for export products. However, the exchange rate relationship
between Japan and South Korea during Abenomics is lower than that in the pre-Abenomics.
Because the prices of the products of South Korea export had a more competitive advantage
than that of Japanese exports, there is a little bit influence in short term during Abenomics,
however, the effect of the depreciation of Japan exchange rate to South Korea exchange
rate is insignificant in long term.
Exchange Market Contagion in Asymmetric Framework before and after Abenomics
67
Table 5: Results of the Enders-Siklos test for threshold co-integration
Entire period
Australia
Canada
China
Euro
Hong Kong
Indonesia
Malaysia
Qatar
Saudi Arabia
Singapore
South Korea
Taiwan
Thailand
United Arab Emirates
U.S.
Pre-Abenomics
During Abenomics
(1)𝑭𝒄
(2)𝑭𝑨
(3)𝛄
(4)𝑭𝒄
(5)𝑭𝑨
(6)𝛄
(7)𝑭𝒄
(8)𝑭𝑨
(9)𝛄
6.378
5.649
2.668
2.570
12.59
3.504
4.441
54.57
10.03
6.133
4.612
3.183
5.609
94.60
3.507
4.418
3.827
0.667
2.646
15.15
3.540
2.682
15.99
5.760
5.132
4.065
3.364
7.089
4.984
4.542
0.0049
-0.0038
6.5e-04
0.0019
-7.3e-05
-0.0051
0.0014
5.5e-05
-2.2e-05
0.0020
-0.0038
0.0015
0.0022
1.4e-05
0.0016
6.217
5.260
1.296
2.383
6.178
0.728
3.951
46.44
34.54
4.097
3.816
3.670
3.813
49.97
3.263
2.179
2.186
1.845
3.020
8.183
1.454
1.457
14.20
4.907
3.374
3.885
2.941
2.572
0.654
4.208
0.0015
-0.0049
1.5e-04
-0.0023
-3.9e-05
5.6e-04
-0.0042
5.7e-05
1.1e-05
0.0020
-0.0038
0.0023
-0.0013
-1.4e-05
-0.0046
5.476
5.464
2.776
1.435
10.79
3.690
3.978
24.84
8.143
5.170
3.641
4.482
4.384
60.87
6.612
5.406
3.285
4.683
1.136
8.907
5.015
1.855
13.27
2.234
4.051
5.110
2.334
5.386
13.40
11.79
0.0024
-0.0037
8.2e-04
-4.1e-04
-6.3e-05
-0.0050
8.5e-04
-1.3e-06
2.2e-05
-0.0015
5.0e-04
0.0013
-0.00275
1.6e-08
-2.0e-04
Co-integration
statistics
(10)=(7)-(4)
Decrease
Increase
Increase
Decrease
Increase
Increase
Increase
Decrease
Decrease
Increase
Decrease
Increase
Increase
Increase
Increase
Asymmetric
Statistics
(11)=(8)-(5)
Increase
Increase
Increase
Decrease
Increase
Increase
Increase
Decrease
Decrease
Increase
Decrease
Decrease
Increase
Increase
Increase
Contagion
Yes/No?
No
Yes
Yes
No
Yes
Yes
Yes
No
No
Yes
No
Yes
Yes
Yes
Yes
Notes:
1. The lag-length of difference Ks selected by minimizing AIC; r is the estimated threshold value.
2. Fc and FA denote the F-statistics for the null hypothesis of no co-integration and symmetric adjustment. Critical values of cointegration test are taken from Enders and Siklos (2001).
3. *, ** and *** denote significance at the 10%, 5% and 1% significance levels, respectively.
68
4
Chien-Chung Nieh and Hsun-Fang Cho
Conclusion
In this paper, we propose the linear models of Engle and Granger (1987) or the Johansen
(1988) co-integration models and threshold co-integration of Enders and Siklos (2001), in
order to investigate financial contagion. This threshold co-integration approach goes
beyond linear co-integration models, analyses the nonlinear adjustment of financial timeseries, and enables us to overcome the estimation error problems of asymmetries. This
enables us to analyze the behavior among exchange rate markets when at least one of them
is under financial crisis (crisis country). To check the robustness of the co-integration
results, we also apply the M-TAR model which can capture an accumulation of changes in
the disequilibrium relationship between variables below and above the threshold followed
by a sharp movement back to the equilibrium position. We successfully apply the threshold
co-integration methodology to the investigation of contagion effect of Japanese yen
devaluation. This study focuses on major trade countries of Japan. To test the existence of
contagion in the exchange rate markets, this paper uses Japan as the crisis country and
Abenomics as triggering off a currency depreciation war. We then estimate the correlations
between the crisis country and all other countries which are surveyed in this study during
both stable and crises periods. Therefore, we split the estimation procedure into subgroups
in order to compare the impact and the magnitude of the spread of the crises in each
individual country. This paper contributes to the literature in the following aspects. First,
we introduce the momentum threshold autoregressive (M-TAR) model of Enders and
Siklos (2001) into the multivariate framework that allows us to capture the correlation
dynamics and asymmetries in a more flexible and realistic way than the linear models of
Engle and Granger (1987) and the Johansen (1988) co-integration models that have been
proposed in the study. Second, we compare the contagion effect of the Japanese yen
depreciation crisis before and after the Abenomics period, and find that most countries with
export trade with Japan follow the depreciation of the yen in applying threshold cointegration model. Third, the empirical evidence confirms a contagion effect particularly in
Asian countries in competition with Japan except South Korea to export. The prices of the
export products of South Korea are far lower and more competitive than that of Japan so
the exchange rate of South Korea is not influenced by the Japanese yen devaluation during
the period of Abenomics. Therefore, the contagion of Japanese yen depreciation does not
transmit to Australia, Euro (France, Germany, Italy, and Netherlands), Qatar, Saudi Arabia,
and the USA with trade competition countries of Japan. We can apparently find out the
effect of yen devaluation just only occurred in Asian area of being close to Japan and does
not spread Europe, America and other countries. Furthermore, Australia is the largest
exporting country of iron ore in the world. However, Japan is the second largest importing
country of iron ore. Depreciation of the yen favored Australian iron ore exports during
Abenomics. In addition, Japan is the third largest importing country of crude petroleum all
over the world. Saudi Arabia is the largest exporting country of crude petroleum.
Devaluation of the yen is an advantage to Saudi Arabia. Qatar also received the same benefit
as Saudi Arabia. In general, our results support the contagion phenomenon for Abenomics.
Nevertheless, the effect of the contagion is regional not global.
Exchange Market Contagion in Asymmetric Framework before and after Abenomics
69
References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
M. E. H. Arouria, M. Bellalahb and D. K. Nguyenc, The co-movements in
international stock markets: new evidence from Latin American emerging Countries,
Applied Economics Letters, 18, (2009), 1 - 6.
T. Baig and I. Goldfajn, Financial market contagion in the Asian crisis, IMF Staff
Papers, 46, (1999).
S. M. Bartram, S. J. Taylor and Y.-H. Wang, The Euro and European financial market
dependence, Journal of Banking & Finance, 31, (2007), 1461 - 1481.
G. Bekaert, C. R. Harvey and A. Ng, Market integration and contagion, Journal of
Business, 78, (2005), 39 - 69.
C. Boucher, Asymmetric adjustment of stock prices to their fundamental value and
the predictability of US stock returns, Economics Letters, 95, (2007), 339 - 347.
S. Calvo and C. M. Reinhart, Capital flows to Latin America: is there evidence of
contagion effects?, In: G. Calvo, M. Goldstein and E. Hochreiter, (Eds.), Private
Capital Flows to Emerging Markets, Institute for International Economics,
Washington, DC, (1996).
G. M. Caporale, A. Cipollini and N. Spagnolo, Testing for Contagion: A Conditional
Correlation Analysis, Journal of Empirical Finance, 12, (2005), 476 - 489.
S. K. Chan, Consistency and limiting distributions of the least squares estimators of a
threshold autoregressive model, Annals of Statistics, 21, (1993), 520 - 533.
S. Chang, Asymmetric Cointegration Relationship among Asian Exchange Rates,
Economic Change and Restructuring, 41, (2008), 125 - 141.
S. Chang, Effects of Asymmetric Adjustment among Labor Productivity, Labor
Demand, and Exchange Rate Using Threshold Cointegration Test, Emerging Markets
Finance and Trade, 46, (2010), 55 - 68.
M. H. Chiang, The Asymmetric Behavior and Spillover Effects on Stock Index
Returns: Evidence on Hong Kong and China, Pan Pacific Management Review, 4,
(2001), 1 - 21.
D. A. Dickey and W. A. Fuller, Likelihood ratio statistics for autoregressive time
series with unit root, Econometric, 49, (1981), 1057 - 1072.
R. Dornbusch, Y. C. Park and S. Claessens, Contagion: Understanding How It
Spreads, The World Bank Research Observer, 15(2), (2000), 177 - 197.
M. Dungey, R. Fry, B. González-Hermosillo and V. Martin, Contagion in
international bond markets during the Russian and the LTCM crises, Journal of
Financial Stability, 2, (2006), 1 - 27.
W. Enders and C. W. J Granger, Unit-root tests and asymmetric adjustment with an
example using the term structure of interest rates, Journal of Business and Economic
Statistics, 16(3), (1998), 304 - 311.
W. Enders and P. L. Siklos, Cointegration and threshold adjustment, Journal of
Business and Economic Statistics, 19(2), (2001), 166 - 176.
R. Engle and S. Yoo, Forecasting and Testing in Co-integration Systems, Journal of
Econometrics, 35, (1987), 143 - 159.
R. F. Engle and C. W. Granger, Co-integration and error correction: Representation,
estimation, and testing, Econometrica, 55, (1987), 251 - 276.
70
Chien-Chung Nieh and Hsun-Fang Cho
[19] K. Forbes and R. Rigobon, Measuring Contagion: Conceptual and Empirical Issues,
The International Bank for Reconstruction and Development organizing conference
of World Bank on International Financial Contagion: How it Spreads and How it Can
Be Stopped, (2001).
[20] K. J. Forbes and R. Rigobon, No contagion, only interdependence: measuring stock
market comovements, Journal of Finance, 57, (2002), 2223 - 2261.
[21] S. Johansen, Statistical Analysis of Cointegration Vectors, Journal of Economic
Dynamics and Control, 12(3), (1988), 231 - 254.
[22] G. Kapetanios, Y. Shin and A. Snell, Testing for a unit root in the nonlinear STAR
framework, Journal of Econometrics, 112, (2003), 359 - 379.
[23] G. Koutmos, Asymmetries in the Conditional Mean and the Conditional Variance:
Evidence from Nine Stock Markets, Journal of Economics and Business, 50, (1998),
277 - 290.
[24] D. Kwiatkowski, P. C. B. Phillips, P. Schmidt and Y. Shin, Testing the Null Hypothesis
of Stationarity against the Alternative of a Unit Root: How Sure are We that Economic
Time Series Have a Unit Root?, Journal of econometrics, 54(1), (1992), 159 - 178.
[25] S. B. Lee and K. J. Kim, Does the October 1987 crash strengthen the co-movements
among national stock markets?, Review of Financial Economics, 3, (1993), 89 - 102.
[26] W. K. Li and K. Lam, Modelling Asymmetry in Stock Returns by a Threshold
Autoregressive Conditional Heteroscedastic Model, The Statistician, 44, (1995), 333
- 341.
[27] M. B. Lucey and S. Voronkova, Russian equity market linkages before and after the
1998 crisis: Evidence from stochastic and regime-switching cointegration tests,
Journal of International Money and Finance, 27, (2008), 1304 - 1324.
[28] S. Ng and P. Perron, Lag length selection and the construction of unit root tests with
good size and power, Econometrica, 69, (2001), 1519 - 1554.
[29] T. Okimoto, New evidence of asymmetric dependence structures in international
equity markets, Journal of Financial and Quantitative Analysis, 43, (2008), 787 - 815.
[30] A. J. Patton, Modelling asymmetric exchange rate dependence, International
Economic Review, 47, (2006), 527 - 556.
[31] P. C. B. Phillips and P. Perron, Testing for a Unit Root in Time Series Regression,
Biometrica, 75(2), (1988), 335 - 346.
[32] J. C. Rodriquez, Measuring financial contagion: a copula approach, Journal of
Empirical Finance, 14, (2007), 401 - 423.
[33] C. H. Shen, C. Chen and L. Chen, An Empirical Study of the Asymmetric
Cointegration Relationships among the Chinese Stock Markets, Applied Economics,
39, (2007), 1433 - 1445.
[34] C. Wang and C. A. Lin, Using Threshold Cointegration to Examine Asymmetric Price
Adjustments between ADR's and Their Underlying Securities—The Case of Taiwan,
South African Journal of Economics, 73, (2005), 449 - 461.